CN110276950B - Urban traffic trip chain reconstruction method based on bayonet video data - Google Patents
Urban traffic trip chain reconstruction method based on bayonet video data Download PDFInfo
- Publication number
- CN110276950B CN110276950B CN201910550349.5A CN201910550349A CN110276950B CN 110276950 B CN110276950 B CN 110276950B CN 201910550349 A CN201910550349 A CN 201910550349A CN 110276950 B CN110276950 B CN 110276950B
- Authority
- CN
- China
- Prior art keywords
- path
- decision
- attribute
- track
- travel time
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/0104—Measuring and analyzing of parameters relative to traffic conditions
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/0104—Measuring and analyzing of parameters relative to traffic conditions
- G08G1/0125—Traffic data processing
- G08G1/0129—Traffic data processing for creating historical data or processing based on historical data
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/017—Detecting movement of traffic to be counted or controlled identifying vehicles
Abstract
The method adopts a PPA path search algorithm (potential path region path search algorithm) to construct an initial decision path, and finally outputs a real path restored by the algorithm by adopting a decision attribute factor model training method; and five decision optimization factors are selected: the path travel time, the path length, the path turning times and the path signal control bayonet number are used as decision attributes, so that the decision factors for track restoration have better environmental adaptability; the decision weight is comprehensively established by the subjective and objective data, so that the method has scientific and practical property and high algorithm speed, and can process large-scale data; the method is suitable for path reconstruction of small and medium-sized road networks, can finish restoration of the missed vehicle track with high precision, has good robustness, and lays a foundation for further statistics of urban traffic road network microscopic parameters.
Description
Technical Field
The invention relates to the field of intelligent transportation, in particular to a method for reconstructing an urban traffic travel chain based on bayonet video data.
Background
With the arrangement of the urban advanced traffic facilities and the continuous change and adjustment of traffic management means and traffic operation models, the traditional resident traffic trip survey has high cost of manpower and financial time, and has low timeliness and accuracy, so that the requirements of new-era traffic planning and management cannot be met.
At present, the development of big data storage and data mining technology in the intelligent transportation field is well-established, and the big data storage and data mining technology is mainly based on traffic basic data, such as Automatic license Plate identification data (ANPR), Global Positioning System (GPS) data, coil data, mobile phone signaling, and the like, and can provide reliable information support for decisions such as traffic policy making, traffic planning and design, traffic control and management, traffic information publishing, and the like by extracting data from a traffic travel track and pushing back Origin-Destination (OD) information and duty ratio information.
The technology adopted by the current traffic control and management based on big data storage and data mining technology has poor accuracy on vehicle travel track restoration, thereby causing certain influence on decisions such as traffic policy making, traffic planning and design, traffic control and management, traffic information publishing and the like.
Disclosure of Invention
In order to solve the defect that the accuracy of vehicle travel track restoration is poor in the prior art, the invention provides a method for reconstructing an urban traffic travel chain based on bayonet video data.
In order to solve the technical problems, the technical scheme of the invention is as follows:
a method for reconstructing an urban traffic trip chain based on bayonet video data comprises the following steps:
step S1: acquiring vehicle license plate data passing through a gate through a video, preprocessing the acquired license plate data, and extracting to obtain a driving track of each vehicle;
step S2: storing paths detected by 4 or more checkpoints continuously in the vehicle trajectory as a historical path data set, and storing data detected by 2 discontinuous checkpoints existing in the vehicle trajectory as the trajectory needing to be restored;
step S3: determining all possible travel track sets T between two checkpoints of track needing restoration according to PPA (potential path area path search algorithm) path algorithmi;
Step S4: calculating a corresponding possible driving track set T in the historical path data setiThe path travel time, the section minimum flow, the path length, the path turning times and the number of signal-controlled bayonets of the path of each path are obtained, and attribute decision factors related to the path travel time, the section minimum flow, the path length, the path turning times and the number of signal-controlled bayonets of the path are obtained;
step S5: all the same path tracks of the starting bayonet and the end point bayonet in the historical path data set are taken out independently, and weight parameters of a reverse-deducing path decision model of the path tracks of each same starting bayonet and end point bayonet are calculated by a method based on an entropy weight Vague set;
step S6: set T of path trajectories and possible travel trajectories with all identical start and end checkpoints in the historical path datasetiAnd taking the path travel time, the path length, the path turning times and the attribute decision factors of the number of signal-controlled bayonets of the path as the input of a path decision model, and combining weight parameters to obtain a final track restoration scheme.
Preferably, the historical path data set in step S2 includes a complete path of the trajectory, a start point bayonet of the trajectory, an end point bayonet of the trajectory, a detected time of the start point of the trajectory, a detected time of the end point of the trajectory, and a travel time of the path.
Preferably, the specific steps of step S3 are as follows:
dividing different road network sub-areas on the whole investigated road network in advance, judging whether the gates belong to the same road network sub-area, if so, carrying out the path search between two points by using a depth-first search algorithm,
if belong to different road network subareasThen, after path search is carried out on the two sub-areas, the two sub-areas are jointed according to the coincident bayonets, and therefore a feasible track set T is obtained preliminarilypre;
Introducing a historical path data set and judging a set TpreWhether each element appears in the historical path dataset or not, and if not, rejecting the set TpreTo obtain a feasible track set Thisto;
The vehicle inlet direction of the end point bayonet point position of the track is restored according to the requirement, and T is further screened outhistoFinally determining possible running track set T of two checkpoints of all to-be-recovered tracksi。
Preferably, in step S4, a set of possible driving trajectories T is calculatediThe specific steps of the attribute decision factors of the path travel time, the minimum flow of the road section, the path length, the path turning times and the number of signal-controlled bayonets of the path are as follows:
defining a travel time matching degree attribute factor GTT(Ti) Road section minimum flow attribute factor GForce(Ti) Path length attribute factor GLen(Ti) Attribute factor G for number of turns of routeC(Ti) Path signal control bayonet quantity attribute factor GI(Ti)5 attribute decision factors, and obtaining G through a standardized evaluation functionTT(Ti)、GForce(Ti)、GLen(Ti)、GC(Ti)、GI(Ti) The normalized evaluation function includes:
(a) a travel time matching degree standardization attribute evaluation function;
(b) a standardized attribute evaluation function of the attraction degree of the traffic cell;
(c) a path length standardized attribute evaluation function;
(d) a path turning number standardization attribute evaluation function;
(e) controlling a bayonet quantity standardized attribute evaluation function by a path signal;
the travel time matching degree standardized attribute evaluation function calculation formula is as follows:
wherein the content of the first and second substances,is TiTheoretical path travel time of (1), tiIs TiThe real travel time of (1); j represents TiEach element in the set, namely one path scheme of reduction, is denoted as scheme j;is the theoretical path travel time, t, in scenario jjIs the true travel time in scenario j;
the calculation formula of the standardized attribute evaluation function of the attraction degree of the traffic cell is as follows:
wherein, FiIs TiAttraction of traffic cells of the approach in the direction of entry, FjThe attraction force of the traffic district of the approach in the direction of the inlet in the scheme j is calculated by using a gravity model according to the accumulation of the attraction force between the adjacent bayonets through which the track passes, and specifically as follows,
wherein QentranceRepresents TiFlow rate of the first section of the route through which the inlet passes, FcarRepresenting an attracted automobile unit, the value is 1, t represents road impedance and is equivalent to travel time required by completing a track;
the path length standardized attribute evaluation function calculation formula is as follows:
wherein L isiIs TiThe path length of (1); l isjFor the path length in scheme j
The path turning number standardized attribute evaluation function calculation formula is as follows:
wherein N isiIs TiThe number of turns required in the process; n is a radical ofjThe number of times of turning is needed in the scheme j;
the path signal control bayonet quantity standardized attribute evaluation function calculation formula is as follows:
wherein, IiIs TiThe number of the middle signal control bayonets; i isjThe number of bayonets is controlled by signals in the scheme j;
initializing a feasible path trajectory set TiTo find GTT(Ti)、GForce(Ti)、GLen(Ti)、GC(Ti)、GI(Ti)。
Preferably, the specific steps of step S5 are as follows:
setting 5 attribute decision factors GTT(Ti)、GForce(Ti)、GLen(Ti)、GC(Ti)、GI(Ti) Weight w of1,w2,w3,w4,w5Determining a decision weight value by training and fitting a historical path data set;
model input in historical path data setWith the same start and end bayonets and TiIs determined by the decision attribute factor GTT(Ti)、GForce(Ti)、GLen(Ti)、GC(Ti)、GI(Ti) Where i is 1, K, n, the model is a set of possible driving trajectories T ═ { T ═ T }1,T2,Ti,K,Tn};
The objective weights of the decision attribute factors are calculated by an entropy weight method, which comprises the following steps,
the entropy of the decision system can be calculated by the formula:calculating to obtain that the smaller the information entropy of the decision attribute is, the larger the provided information is, the larger the weight distributed by the decision attribute factor should be, and otherwise, the smaller the distributed weight should be;
after the decision attribute entropy value is compensated, the objective weight of the decision attribute obtained by normalization processing is as follows:
the decision attribute objective weight vector is thus: w ═ w1',w'2,K,w'5)
Subjective decision attribute weight w ═ (w ═1”,w'2',K,w'5') to obtain a decision attribute weight vector interval as:
let W ═ ([ W ]11,w12],[w21,w22],K,[w51,w52])。
Preferably, the specific steps of step S6 are as follows:
let d be (d)1,d2,K,d5)TVector representing decision attribute, diThe attribute factor standard value of each scheme under the decision attribute i is included:
Sj={di∈d|λij>λU}
Oj={di∈d|λij<λL}
Nj={di∈d|λL≤λij≤λU}
Sj,Oj,Njrespectively, a decision support set of the scheme j, a decision objection set of the scheme j and a decision neutral set of the scheme j;
for alternative T ═ T1,T2,Ti,K,TnFor the degree to which it meets the requirements on 5 decision attributes, with a value V of valuet∈TWhere t represents each alternative, Vt∈T=[lt∈T,CIdt∈T]I is a total set of weights, i.e. each element is 1, lt∈T=[αt∈T,βt∈T],dt∈T=[μt∈T,ηt∈T]
Defining a decision objective function:
The final trajectory restoration scheme is: max (f)t∈T)。
Compared with the prior art, the technical scheme of the invention has the beneficial effects that:
the method adopts a PPA algorithm to construct an initial decision path, and adopts a decision attribute factor model training method to finally output a real path restored by the algorithm; and five decision optimization factors are selected: the path travel time, the path length, the path turning times and the path signal control bayonet number are used as decision attributes, so that the decision factors for track restoration have better environmental adaptability; the decision weight is comprehensively established by the subjective and objective data, so that the method has scientific and practical property and high algorithm speed, and can process large-scale data; the method is suitable for path reconstruction of small and medium-sized road networks, can finish restoration of the missed vehicle track with high precision, has good robustness, and lays a foundation for further statistics of urban traffic road network microscopic parameters.
Drawings
FIG. 1 is a flow chart of the method of the present invention.
Fig. 2 is a PPA path search diagram according to a second embodiment of the present invention.
Fig. 3 is a schematic diagram of an entropy weight decision process according to a second embodiment of the present invention.
Detailed Description
The drawings are for illustrative purposes only and are not to be construed as limiting the patent;
for the purpose of better illustrating the embodiments, certain features of the drawings may be omitted, enlarged or reduced, and do not represent the size of an actual product;
it will be understood by those skilled in the art that certain well-known structures in the drawings and descriptions thereof may be omitted.
The technical solution of the present invention is further described below with reference to the accompanying drawings and examples.
Example 1
As shown in fig. 1, a method for reconstructing an urban transportation trip chain based on bayonet video data includes the following steps:
step S1: acquiring vehicle license plate data passing through a gate through a video, preprocessing the acquired license plate data, and extracting to obtain a driving track of each vehicle;
step S2: storing paths detected by 4 or more checkpoints continuously in the vehicle trajectory as a historical path data set, and storing data detected by 2 discontinuous checkpoints existing in the vehicle trajectory as the trajectory needing to be restored;
step S3: determining all possible travel track sets T between two bayonets of the track needing to be restored according to the PPA path algorithmi;
Step S4: calculating a corresponding possible driving track set T in the historical path data setiThe path travel time, the section minimum flow, the path length, the path turning times and the number of signal-controlled bayonets of the path of each path are obtained, and attribute decision factors related to the path travel time, the section minimum flow, the path length, the path turning times and the number of signal-controlled bayonets of the path are obtained;
step S5: all the same path tracks of the starting bayonet and the end point bayonet in the historical path data set are taken out independently, and weight parameters of a reverse-deducing path decision model of the path tracks of each same starting bayonet and end point bayonet are calculated by a method based on an entropy weight Vague set;
step S6: set T of path trajectories and possible travel trajectories with all identical start and end checkpoints in the historical path datasetiAnd taking the path travel time, the path length, the path turning times and the attribute decision factors of the number of signal-controlled bayonets of the path as the input of a path decision model, and combining weight parameters to obtain a final track restoration scheme.
As a preferred embodiment, the historical path data set described in step S2 includes a track complete path, a track start point bayonet, a track end point bayonet, a track start point detected time, a track end point detected time, and a path travel time.
As a preferred embodiment, the specific steps of step S3 are as follows:
dividing different road network sub-areas on the whole investigated road network in advance, judging whether the gates belong to the same road network sub-area, if so, carrying out the path search between two points by using a depth-first search algorithm,
if the two subzones belong to different road network subzones, after path search is carried out on the two subzones, the two subzones are jointed according to the superposed bayonets, and therefore a feasible rail is obtained preliminarilyTrace set Tpre;
Introducing a historical path data set and judging a set TpreWhether each element appears in the historical path dataset or not, and if not, rejecting the set TpreTo obtain a feasible track set Thisto;
The vehicle inlet direction of the end point bayonet point position of the track is restored according to the requirement, and T is further screened outhistoFinally determining possible running track set T of two checkpoints of all to-be-recovered tracksi。
As a preferred embodiment, the set of possible travel trajectories T is calculated in step S4iThe specific steps of the attribute decision factors of the path travel time, the minimum flow of the road section, the path length, the path turning times and the number of signal-controlled bayonets of the path are as follows:
defining a travel time matching degree attribute factor GTT(Ti) Road section minimum flow attribute factor GForce(Ti) Path length attribute factor GLen(Ti) Attribute factor G for number of turns of routeC(Ti) Path signal control bayonet quantity attribute factor GI(Ti)5 attribute decision factors, and obtaining G through a standardized evaluation functionTT(Ti)、GForce(Ti)、GLen(Ti)、GC(Ti)、GI(Ti) The normalized evaluation function includes:
(a) a travel time matching degree standardization attribute evaluation function;
(b) a standardized attribute evaluation function of the attraction degree of the traffic cell;
(c) a path length standardized attribute evaluation function;
(d) a path turning number standardization attribute evaluation function;
(e) controlling a bayonet quantity standardized attribute evaluation function by a path signal;
the travel time matching degree standardized attribute evaluation function calculation formula is as follows:
wherein the content of the first and second substances,is TiTheoretical path travel time of (1), tiIs TiThe real travel time of (1); j represents TiEach element in the set, namely one path scheme of reduction, is denoted as scheme j;is the theoretical path travel time, t, in scenario jjIs the true travel time in scenario j;
the calculation formula of the standardized attribute evaluation function of the attraction degree of the traffic cell is as follows:
wherein, FiIs TiAttraction of traffic cells of the approach in the direction of entry, FjThe attraction force of the traffic district of the approach in the direction of the inlet in the scheme j is calculated by using a gravity model according to the accumulation of the attraction force between the adjacent bayonets through which the track passes, and specifically as follows,
wherein QentranceRepresents TiFlow rate of the first section of the route through which the inlet passes, FcarRepresenting an attracted automobile unit, the value is 1, t represents road impedance and is equivalent to travel time required by completing a track;
the path length standardized attribute evaluation function calculation formula is as follows:
wherein L isiIs TiThe path length of (1); l isjFor the path length in scheme j
The path turning number standardized attribute evaluation function calculation formula is as follows:
wherein N isiIs TiThe number of turns required in the process; n is a radical ofjThe number of times of turning is needed in the scheme j;
the path signal control bayonet quantity standardized attribute evaluation function calculation formula is as follows:
wherein, IiIs TiThe number of the middle signal control bayonets; i isjThe number of bayonets is controlled by signals in the scheme j;
initializing a feasible path trajectory set TiTo find GTT(Ti)、GForce(Ti)、GLen(Ti)、GC(Ti)、GI(Ti)。
As a preferred embodiment, the specific steps of step S5 are as follows:
setting 5 attribute decision factors GTT(Ti)、GForce(Ti)、GLen(Ti)、GC(Ti)、GI(Ti) Weight w of1,w2,w3,w4,w5Determining a decision weight value by training and fitting a historical path data set;
the model is input with all the same start and end point bayonets and T in the historical path datasetiIn (c) blockPolicy attribute factor GTT(Ti)、GForce(Ti)、GLen(Ti)、GC(Ti)、GI(Ti) Where i is 1, K, n, the model is a set of possible driving trajectories T ═ { T ═ T }1,T2,Ti,K,Tn};
The objective weights of the decision attribute factors are calculated by an entropy weight method, which comprises the following steps,
the entropy of the decision system can be calculated by the formula:calculating to obtain that the smaller the information entropy of the decision attribute is, the larger the provided information is, the larger the weight distributed by the decision attribute factor should be, and otherwise, the smaller the distributed weight should be;
after the decision attribute entropy value is compensated, the objective weight of the decision attribute obtained by normalization processing is as follows:
the decision attribute objective weight vector is thus: w ═ w1',w'2,K,w'5)
Subjective decision attribute weight w ═ w "1,w”2,K,w”5) Thus, the obtained decision attribute weight vector interval is:
let W ═ ([ W ]11,w12],[w21,w22],K,[w51,w52])。
As a preferred embodiment, the specific steps of step S6 are as follows:
let d be (d)1,d2,K,d5)TVector representing decision attribute, diThe attribute factor standard value of each scheme under the decision attribute i is included:
Sj={di∈d|λij>λU}
Oj={di∈d|λij<λL}
Nj={di∈d|λL≤λij≤λU}
Sj,Oj,Njrespectively, a decision support set of the scheme j, a decision objection set of the scheme j and a decision neutral set of the scheme j;
for alternative T ═ T1,T2,Ti,K,TnFor the degree to which it meets the requirements on 5 decision attributes, with a value V of valuet∈TWhere t represents each alternative, Vt∈T=[lt∈T,CIdt∈T]I is a total set of weights, i.e. each element is 1, lt∈T=[αt∈T,βt∈T],dt∈T=[μt∈T,ηt∈T]
Defining a decision objective function:
The final trajectory restoration scheme is: max (f)t∈T)。
Example 2
As shown in fig. 1, 2 and 3, in the present embodiment, data statistics is performed every 30min for 24 hours a day, and after dividing, 48 time statistics windows can be obtained
The collected data are preprocessed according to the statistical time windows to obtain sampled travel time samples, and the estimation method of the travel time of the road section under each statistical time window is as follows,
firstly, noise processing is respectively carried out on the conditions of different traffic jams in one day by counting road section travel time data collected under a time window;
firstly, for the road section travel time of different time periods, T is set according to the condition that the travel time at the moment is larger than the free flow timel=Tfreeflow*α,TfreeflowIn order to give a certain threshold value space to the overspeed sample of the urban road for the free flow time of the road section, alpha is generally 0.8-0.9, and the sufficiency of sampling is ensured;
for peak time (the coming and disappearing of the peak time can be judged according to the historical data of specific urban traffic, the early peak is generally determined to be 7:00-9:00 or 7:30:00-9:30:00, the late peak is generally determined to be 17:00:00-19:00 or 16:30:00-18:30:00), and the elimination travel time exceeds TuSample of (2), TuThe upper limit of the road section travel time, which may be 1800s here, is listed as noise data;
for the flat peak period (generally determined to be 10:00:00-16:00:00), TuRetrievable Tu=β*TfreeflowBeta is about 3.5 generally, the threshold value is adjusted according to the road section characteristics, the road congestion level of the urban road network in the peak leveling period is in a medium or smooth state under the normal traffic operation condition, so that the phenomenon of short stop of the vehicle between two gate points possibly occurs for individual large travel time samples;
for the evening early morning hours (0:00:00-6:00:00, 19:00:00-23:59:59), TuRetrievable Tu=γ*TfreeflowGamma is about 1.5, and the influence of noise samples in the period is eliminated;
obtaining travel time sample data after the preliminary screening by the steps, calculating outliers in a statistical time window, setting appropriate parameters according to requirements to eliminate outlier noise samples,
TT”'N=(TT”Ni|TT'N.mean-2σN.mean≤TT”Ni≤TT'N.mean+2σN.mean)
TT”'N=(TT”Ni|TT”N.median-3DNf.mean≤TT”'Ni≤TT”N.median+3DNa.mean)
wherein TT'N.medianIs TT'NThe median of (a) is determined,
(in the case where n is an even number, m is n/2, odd, and so on) is the average absolute deviation of the sample from the median of the sample;
repeating the steps for multiple times according to the actual data condition until the noise data are eliminated completely, and ensuring the clustering and representativeness of the researched travel time data;
corresponding the initial time corresponding to the cleaned travel time vehicle sample to a statistical time window, averaging all travel times of the statistical time window to obtain a road section travel time estimated value in the statistical time window, and recording the road section travel time estimated values of all the statistical time windows;
calculating to obtain travel time estimated values of all statistical time windows of all road sections, and storing the travel time estimated values for calculating the path theoretical travel time matched with the subsequent travel time;
preprocessing the acquired data according to a statistical time window to obtain the vehicle flow of each flow direction of each road section;
preliminarily extracting a travel path according to the urban road grade, the vehicle speed limit, the road section maximum travel time threshold and the road topological structure;
storing paths continuously detected by 4 or more bayonets as a historical path data set, wherein the data set comprises a complete path of a track, a bayonet of a starting point of the track, a bayonet of an end point of the track, detected time of the starting point of the track, detected time of the end point of the track and travel time of the path;
recording data detected by two discontinuous bayonet detection points in the driving track of each vehicle, and marking the data as a track segment needing to be restored;
determining all possible travel track sets between two checkpoints needing to restore the track (namely, paths which are not detected by adjacent checkpoint equipment in the road network topology);
suppose that fig. 2 shows a road network of an urban road network, the road network comprises 25 nodes and 35 road segments, and 2 path search sub-areas are divided by the road network;
if two discontinuous bayonet detection points in the road network in fig. 2 are '6' and '22', a feasible path can be determined according to the PPA path search algorithm, and the specific steps are as follows,
dividing different road network sub-areas on the whole investigated road network in advance, judging whether the bayonets belong to the same road network sub-area, and if the bayonets belong to the same road network sub-area, performing path search between two points by using a depth-first search (DFS) algorithm;
if the intersection belongs to different road network sub-areas, path searching and jointing are carried out according to the intersection with the two overlapped sub-areas;
if the feasible track set T belongs to different road network sub-areas, after path search is carried out on the two sub-areas, path bonding is carried out according to the connecting bayonets, and therefore a feasible track set T is obtainedpre={T1,T2,K,TkK is the total number of the prepared tracks meeting the condition of the recovery tracks;
introducing a historical path data set, and judging a set TpreWhether each element appears in the historical path dataset or not, and if not, rejecting the set TpreTo obtain a feasible track set Thisto={T1,T2,K,Tm};
According to the vehicle inlet direction of the end point bayonet point of the track to be restored, the generated feasible road sets can be further screened out, and finally the possible running track sets T ═ T of the two bayonets of the track to be restored are determined1,T2,Ti,K,TnN is the total number of the preparation tracks after the preliminary screening;
setting travel time matching degree attribute factors, section minimum flow attribute factors, path length attribute factors, path turning times attribute factors and path signal control bayonet quantity attribute factors, namely 5 attribute decision factors, and designing a standardized evaluation function:
travel time matching degree standardization attribute evaluation function:
wherein the content of the first and second substances,is TiTheoretical path travel time of (1), tiIs TiThe real travel time of (1);
standardized attribute evaluation function of the attraction degree of the traffic cell:
wherein T is TiSet of links of the route in (1), FiIs TiThe attractive force of the traffic district of the approach in the inlet direction is calculated by using a gravity model according to the accumulation of the attractive force between the adjacent bayonets through which the track passes, and the attractive force is calculated as follows,
wherein QentranceRepresents TiFlow rate of the first section of the route through which the inlet passes, FcarRepresenting an attracted automobile unit, the value is 1, t represents road impedance and is equivalent to travel time required by completing a track;
path length normalized attribute evaluation function:
wherein L isiIs TiIn (1)A path length;
path turn number normalized attribute evaluation function:
wherein N isiIs TiThe number of turns required in the process;
path signal control bayonet quantity standardization attribute evaluation function:
wherein, IiIs TiThe number of the middle signal control bayonets;
initializing a feasible path trajectory set TiTo find GTT(Ti)、GForce(Ti)、GLen(Ti)、GC(Ti)、GI(Ti)。
According to the historical path data set, all path set records corresponding to all the same starting bayonets and all the same end bayonets are extracted independently, and historical path set data of all the same starting bayonets and all the same end bayonets can be used for subsequent training of attribute factor decision parameters;
setting 5 attribute decision factors GTT(Ti)、GForce(Ti)、GLen(Ti)、GC(Ti)、GI(Ti) Weight w of1,w2,w3,w4,w5The decision weight value is obtained by performing training fitting according to a historical data set;
calculating to obtain the weight parameter of each reverse-thrust path decision model of all the same start bayonets and end bayonets by a method based on an entropy weight Vague set, wherein the input of the model is the separately stored T in all the same history track sets between the start bayonets and the end bayonetsiG is a decision attribute factorTT(Ti)、GForce(Ti)、GLen(Ti)、GC(Ti)、GI(Ti) Where i is 1, K, n, as shown in fig. 3, the model is a set of possible driving trajectories T ═ { T ═ T { (T) } in the set of possible driving trajectories1,T2,Ti,K,Tn};
The objective weights of the decision attribute factors are calculated by an entropy weight method, which comprises the following steps,
the entropy of the decision system can be calculated by the formula:calculating to obtain that the smaller the information entropy of the decision attribute is, the larger the provided information is, the larger the weight distributed by the decision attribute factor should be, and otherwise, the smaller the distributed weight should be;
after the decision attribute entropy value is compensated, the objective weight of the decision attribute obtained by normalization processing is as follows:
the decision attribute objective weight vector is thus: w ═ w1',w'2,K,w'5)
Subjective decision attribute weight w ═ w "1,w”2,K,w”5) Thus, the obtained decision attribute weight vector interval is:
let W ═ ([ W ]11,w12],[w21,w22],K,[w51,w52]);
Let d be (d)1,d2,K,d5)TVector representing decision attribute, diThe attribute factor standard value of each scheme under the decision attribute i is included:
Sj={di∈d|λij>λU}
Oj={di∈d|λij<λL}
Nj={di∈d|λL≤λij≤λU}
Sj,Oj,Njrespectively, a decision support set of the scheme j, a decision objection set of the scheme j and a decision neutral set of the scheme j;
for alternative T ═ T1,T2,Ti,K,TnFor the degree to which it meets the requirements on 5 decision attributes, with a value V of valuet∈TWhere t represents each alternative, Vt∈T=[lt∈T,CIdt∈T]I is a total set of weights, i.e. each element is 1, lt∈T=[αt∈T,βt∈T],dt∈T=[μt∈T,ηt∈T]
Defining a decision objective function:
The final trajectory restoration scheme is: max (f)t∈T)。
The same or similar reference numerals correspond to the same or similar parts;
the terms describing positional relationships in the drawings are for illustrative purposes only and are not to be construed as limiting the patent;
it should be understood that the above-described embodiments of the present invention are merely examples for clearly illustrating the present invention, and are not intended to limit the embodiments of the present invention. Other variations and modifications will be apparent to persons skilled in the art in light of the above description. And are neither required nor exhaustive of all embodiments. Any modification, equivalent replacement, and improvement made within the spirit and principle of the present invention should be included in the protection scope of the claims of the present invention.
Claims (1)
1. A method for reconstructing an urban traffic trip chain based on bayonet video data is characterized by comprising the following steps:
step S1: collecting video data of vehicles passing through the gate, processing the video data to obtain license plate data of the vehicles passing through the gate, and constructing a driving track of each vehicle based on the vehicle license plate data extracted from different gates;
step S2: storing paths detected by 4 or more checkpoints continuously in the driving track of each vehicle as a historical path data set, and marking the paths with missed detection of the checkpoints in the driving track of each vehicle as the tracks needing to be restored;
step S3: determining all possible travel track sets T between the head and the tail of the tracks needing to be restored according to a PPA path algorithmi;
Step S4: initializing a set of possible trajectories TiSet of possible travel trajectories TiObtaining attribute decision factors related to the path travel time, the section minimum flow, the path length, the path turning times and the number of signal-controlled intersections of the path;
step S5: constructing a path decision model, taking out paths with the same head and tail bayonets as those of a track to be restored in a historical path data set, and calculating a reverse-deducing by a method based on an entropy weight Vague set based on the data to obtain a weight parameter of the path decision model;
step S6: with possible travel trajectory set TiThe two bayonets at the head and the tail of the track needing to be recovered are the sameTaking the path travel time, the path length, the path turning times of the path and the attribute decision factors and weight parameters of the number of intersections of the path controlled by the signal as the input of a path decision model to obtain a final track restoration scheme;
the historical path data set in the step S2 includes a complete path of the trajectory, a start point bayonet of the trajectory, an end point bayonet of the trajectory, detected time of the start point of the trajectory, detected time of the end point of the trajectory, and travel time of the path;
the specific steps of step S3 are as follows:
dividing the road section provided with the bayonets into different road network sub-areas, judging whether the bayonets between two pairs belong to the same road network sub-area, if the bayonets belong to the same road network sub-area, performing path search between two points by using a depth-first search algorithm, if the bayonets belong to different road network sub-areas, performing path search in the two road network sub-areas, and then jointing according to overlapped bayonets, thereby preliminarily obtaining a feasible track set Tpre;
Introducing a historical path data set and judging a set TpreWhether each element appears in the historical path dataset or not, and if not, rejecting the set TpreTo obtain a feasible track set Thisto;
The vehicle inlet direction of the end point bayonet point position of the track is restored according to the requirement, and T is further screened outhistoFinally determining all possible running track sets T between the head and the tail of the two checkpoints in all recovered tracksi;
In step S4, a set of possible driving trajectories T is calculatediThe specific steps of the attribute decision factors of the path travel time, the minimum flow of the road section, the path length, the path turning times and the number of signal-controlled bayonets of the path are as follows:
defining a travel time matching degree attribute factor GTT(Ti) Road section minimum flow attribute factor GForce(Ti) Path length attribute factor GLen(Ti) Attribute factor G for number of turns of routeC(Ti) Route informationNumber control bayonet quantity attribute factor GI(Ti)5 attribute decision factors, and obtaining G through a standardized evaluation functionTT(Ti)、GForce(Ti)、GLen(Ti)、GC(Ti)、GI(Ti) The normalized evaluation function includes:
(a) a travel time matching degree standardization attribute evaluation function;
(b) a standardized attribute evaluation function of the attraction degree of the traffic cell;
(c) a path length standardized attribute evaluation function;
(d) a path turning number standardization attribute evaluation function;
(e) controlling a bayonet quantity standardized attribute evaluation function by a path signal;
the travel time matching degree standardized attribute evaluation function calculation formula is as follows:
wherein the content of the first and second substances,is TiTheoretical path travel time of (1), tiIs TiThe real travel time of (1); j represents TiEach element in the set, namely one path scheme of reduction, is denoted as scheme j;is the theoretical path travel time, t, in scenario jjIs the true travel time in scenario j;
the calculation formula of the standardized attribute evaluation function of the attraction degree of the traffic cell is as follows:
wherein, FiIs TiAttraction of traffic cells of the approach in the direction of entry, FjThe attraction force of the traffic district of the approach in the direction of the inlet in the scheme j is calculated by using a gravity model according to the accumulation of the attraction force between the adjacent bayonets through which the track passes, and specifically as follows,
wherein QentranceRepresents TiFlow rate of the first section of the route through which the inlet passes, FcarRepresenting an attracted automobile unit, the value is 1, t represents road impedance and is equivalent to travel time required by completing a track;
the path length standardized attribute evaluation function calculation formula is as follows:
wherein L isiIs TiThe path length of (1); l isjFor the path length in scheme j
The path turning number standardized attribute evaluation function calculation formula is as follows:
wherein N isiIs TiThe number of turns required in the process; n is a radical ofjThe number of times of turning is needed in the scheme j;
the path signal control bayonet quantity standardized attribute evaluation function calculation formula is as follows:
wherein, IiIs TiThe number of the middle signal control bayonets;Ijthe number of bayonets is controlled by signals in the scheme j;
initializing a feasible path trajectory set TiTo find GTT(Ti)、GForce(Ti)、GLen(Ti)、GC(Ti)、GI(Ti);
The specific steps of calculating the weight parameter of the path decision model in step S5 are as follows:
defining 5 attribute decision factors GTT(Ti)、GForce(Ti)、GLen(Ti)、GC(Ti)、GI(Ti) Has a weight of w1,w2,w3,w4,w5Determining a decision weight value by training and fitting a historical path data set;
input of model by TiIs determined by the decision attribute factor GTT(Ti)、GForce(Ti)、GLen(Ti)、GC(Ti)、GI(Ti) Where i 1., n, the model is a set of possible driving trajectories T ═ T ·1,T2,Ti,...,Tn};
The objective weights of the decision attribute factors are calculated by an entropy weight method, which comprises the following steps,
order toλijWhere i is a decision factor G of 5 attributesTT(Ti)、GForce(Ti)、GLen(Ti)、GC(Ti) J is TiN possible driving trajectory plans;
the entropy of the decision system can be calculated by the formula:
wherein the content of the first and second substances,the smaller the information entropy of the decision attribute is, the larger the provided information is, the larger the weight distributed by the decision attribute factor should be, otherwise, the smaller the distributed weight should be;
after the decision attribute entropy value is compensated, the objective weight of the decision attribute obtained by normalization processing is as follows:
thereby obtaining 5 attribute decision factors GTT(Ti)、GForce(Ti)、GLen(Ti)、GC(Ti)、GI(Ti) The decision attribute objective weight vector of (1) is: w ═ w'1,w′2,...,w′5)
Let 5 attribute decision factors GTT(Ti)、GForce(Ti)、GLen(Ti)、GC(Ti)、GI(Ti) The subjective decision attribute weight of (a) is w ″ ═ w ″1,w″2,...,w″5) Thus, the obtained decision attribute weight vector interval is:
let W ═ ([ W ]11,w12],[w21,w22],...,[w51,w52]);
The specific steps of obtaining the final trajectory restoration scheme in step S6 are as follows:
definition d ═ (d)1,d2,...,d5)TVector representing decision attribute, diThe method comprises the following steps of including attribute factor standard values of each possible driving track under decision attributes:
Sj={di∈d|λij>λU}
Oj={di∈d|λij<λL}
Nj={di∈d|λL≤λij≤λU}
Sj,Oj,Njrespectively defining a decision support set of a scheme j, a decision objection set of the scheme j and a decision neutral set of the scheme j;
for T ═ T1,T2,Ti,...,TnIt is at G }TT(Ti)、GForce(Ti)、GLen(Ti)、GC(Ti)、GI(Ti) The figure value V is used to the extent that the 5 decision attributes satisfy the requirementst∈TWhere t represents each alternative, Vt∈T=[lt∈T,CIdt∈T]I is a total set of weights, i.e. each element is 1, lt∈T=[αt∈T,βt∈T],dt∈T=[μt∈T,ηt∈T];
Defining a decision objective function:
The final trajectory restoration scheme is: max (f)t∈T)。
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910550349.5A CN110276950B (en) | 2019-06-24 | 2019-06-24 | Urban traffic trip chain reconstruction method based on bayonet video data |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910550349.5A CN110276950B (en) | 2019-06-24 | 2019-06-24 | Urban traffic trip chain reconstruction method based on bayonet video data |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110276950A CN110276950A (en) | 2019-09-24 |
CN110276950B true CN110276950B (en) | 2021-12-31 |
Family
ID=67961789
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910550349.5A Active CN110276950B (en) | 2019-06-24 | 2019-06-24 | Urban traffic trip chain reconstruction method based on bayonet video data |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110276950B (en) |
Families Citing this family (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113763696B (en) * | 2020-06-01 | 2023-05-02 | 杭州海康威视数字技术股份有限公司 | Vehicle path reconstruction method and device, electronic equipment and storage medium |
CN111768619A (en) * | 2020-06-16 | 2020-10-13 | 苏州大学 | Express way vehicle OD point determining method based on checkpoint data |
CN112215427B (en) * | 2020-10-19 | 2022-12-23 | 山东交通学院 | Vehicle driving track reconstruction method and system under condition of bayonet data loss |
CN112365711B (en) * | 2020-10-21 | 2021-11-02 | 东南大学 | Vehicle track reconstruction method based on license plate recognition data |
CN112289026B (en) * | 2020-10-26 | 2022-01-04 | 山东旗帜信息有限公司 | Vehicle route restoration method, equipment and medium |
CN112309118B (en) * | 2020-11-03 | 2021-11-09 | 广州市交通规划研究院 | Vehicle trajectory calculation method based on space-time similarity |
CN112985442B (en) * | 2021-03-03 | 2022-11-04 | 北京嘀嘀无限科技发展有限公司 | Driving path matching method, readable storage medium and electronic device |
CN113129589B (en) * | 2021-03-26 | 2022-08-16 | 中山大学 | Individual OD cell inference method based on bayonet detection data |
CN113297342B (en) * | 2021-05-18 | 2022-05-10 | 北京理工大学前沿技术研究院 | Vehicle driving track reconstruction method, device, equipment and storage medium |
CN115512543B (en) * | 2022-09-21 | 2023-11-28 | 浙江大学 | Vehicle path chain reconstruction method based on deep reverse reinforcement learning |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103440764A (en) * | 2013-08-19 | 2013-12-11 | 同济大学 | Urban road network vehicle travel path reconstruction method based on vehicle automatic identification data |
CN104778274A (en) * | 2015-04-23 | 2015-07-15 | 山东大学 | Wide-range urban road network travel time estimation method based on sparse taxi GPS (Global Positioning System) data |
CN105606110A (en) * | 2015-11-03 | 2016-05-25 | 中兴软创科技股份有限公司 | Depth-first traversal-based feasible path searching method and device |
CN105678410A (en) * | 2015-12-31 | 2016-06-15 | 哈尔滨工业大学 | Public traffic system space-time reachability modeling method of considering network connectivity time varying characteristics |
CN106023589A (en) * | 2016-06-16 | 2016-10-12 | 北京航空航天大学 | Gate data-based vehicle trajectory reconstruction method |
CN107195180A (en) * | 2017-06-08 | 2017-09-22 | 青岛海信网络科技股份有限公司 | A kind of traffic trip track extraction method and device based on the alert data of electricity |
CN108335485A (en) * | 2018-01-31 | 2018-07-27 | 夏莹杰 | The method of major issue traffic dynamic emulation congestion prediction based on license plate identification data |
CN109166309A (en) * | 2018-08-06 | 2019-01-08 | 重庆邮电大学 | A kind of missing data on flows restoration methods towards complicated urban traffic network |
Family Cites Families (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108133611A (en) * | 2016-12-01 | 2018-06-08 | 中兴通讯股份有限公司 | Vehicle driving trace monitoring method and system |
CN108961747B (en) * | 2018-07-03 | 2019-11-05 | 北京航空航天大学 | A kind of urban road traffic state information extracting method under incomplete bayonet data qualification |
CN109190056B (en) * | 2018-08-22 | 2020-07-24 | 深圳先进技术研究院 | Vehicle track reconstruction method and system and electronic equipment |
-
2019
- 2019-06-24 CN CN201910550349.5A patent/CN110276950B/en active Active
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103440764A (en) * | 2013-08-19 | 2013-12-11 | 同济大学 | Urban road network vehicle travel path reconstruction method based on vehicle automatic identification data |
CN104778274A (en) * | 2015-04-23 | 2015-07-15 | 山东大学 | Wide-range urban road network travel time estimation method based on sparse taxi GPS (Global Positioning System) data |
CN105606110A (en) * | 2015-11-03 | 2016-05-25 | 中兴软创科技股份有限公司 | Depth-first traversal-based feasible path searching method and device |
CN105678410A (en) * | 2015-12-31 | 2016-06-15 | 哈尔滨工业大学 | Public traffic system space-time reachability modeling method of considering network connectivity time varying characteristics |
CN106023589A (en) * | 2016-06-16 | 2016-10-12 | 北京航空航天大学 | Gate data-based vehicle trajectory reconstruction method |
CN107195180A (en) * | 2017-06-08 | 2017-09-22 | 青岛海信网络科技股份有限公司 | A kind of traffic trip track extraction method and device based on the alert data of electricity |
CN108335485A (en) * | 2018-01-31 | 2018-07-27 | 夏莹杰 | The method of major issue traffic dynamic emulation congestion prediction based on license plate identification data |
CN109166309A (en) * | 2018-08-06 | 2019-01-08 | 重庆邮电大学 | A kind of missing data on flows restoration methods towards complicated urban traffic network |
Non-Patent Citations (2)
Title |
---|
基于熵权Vague集的多目标决策方法;赵庆庆;《计算机应用》;20180510;1250-1253 * |
适用于不同交通状态的交通控制小区动态划分方法研究;曾令宇;《中国优秀硕士学位论文全文数据库工程科技II辑》;20180715;C034-398 * |
Also Published As
Publication number | Publication date |
---|---|
CN110276950A (en) | 2019-09-24 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110276950B (en) | Urban traffic trip chain reconstruction method based on bayonet video data | |
CN109410577B (en) | Self-adaptive traffic control subarea division method based on space data mining | |
CN112365711B (en) | Vehicle track reconstruction method based on license plate recognition data | |
CN113724504B (en) | Urban area traffic prediction system and method oriented to vehicle track big data | |
CN104715610B (en) | A kind of traffic index computational methods towards urban transportation | |
CN102087788B (en) | Method for estimating traffic state parameter based on confidence of speed of float car | |
CN104778834B (en) | Urban road traffic jam judging method based on vehicle GPS data | |
CN111210612B (en) | Method for extracting bus route track based on bus GPS data and station information | |
CN107016851A (en) | The method that a kind of quantitative analysis city built environment influences on road journey time | |
CN103984994B (en) | Method for predicting urban rail transit passenger flow peak duration | |
CN108648445B (en) | Dynamic traffic situation prediction method based on traffic big data | |
CN109345832B (en) | Urban road overtaking prediction method based on deep recurrent neural network | |
CN100466010C (en) | Different species traffic information real time integrating method | |
CN107180270A (en) | Passenger flow forecasting and system | |
CN112784000B (en) | Passenger searching method based on taxi track data | |
CN109275121A (en) | A kind of Vehicle tracing method based on adaptive extended kalman filtering | |
CN106251642A (en) | A kind of public transport road based on real-time bus gps data chain speed calculation method | |
CN112862240B (en) | Road obstacle risk assessment method and device based on urban big data and readable storage medium | |
CN113537626B (en) | Method for predicting neural network combined time sequence by aggregating information difference | |
CN113868492A (en) | Visual OD (origin-destination) analysis method based on electric police and checkpoint data and application | |
CN110021161B (en) | Traffic flow direction prediction method and system | |
CN116913088A (en) | Intelligent flow prediction method for expressway | |
CN116227362A (en) | Municipal drainage pipe network state prediction method based on graph and deep learning | |
CN107730882A (en) | Congestion in road forecasting system and method based on artificial intelligence | |
CN110211380B (en) | Multisource traffic data fusion expressway congestion interval detection method |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |